Research Of Sorting Technology Based On Industrial Robot Of Machine Vision

With the vigorous development of technology, robot technology is advancing rapidly.People turn the focus on robot vision technology field, making the robot have similar to thefunction of human’s eye. After decades of development, machine vision technology haveplayed an important role in food packaging, pharmaceutical production, non-destructivetesting, precision control, automatic production lines, as well as all kinds of dangerousmanual occasions. However there are not good solution to complexity of the machinevision technology and uncertain environment, it is an important direction of research to thedevelopment of visual robot in the future.Aimed at the workpiece sorting problem on industrial production line, usingMicrosoft Visual C++integrated development environment have set up image processingalgorithm library, which is including the camera calibration interactive interface, patternrecognition of sorting function algorithms and dynamic workpiece tracking algorithm, andintroduce to solve the related methods of problem from the two respect of static anddynamic workpiece sorting. Against sorting issues of static workpiece placed on theplatform, This article not only analyzes camera model in the OpenCV vision algorithmslibrary, calibrate monocular camera with method of Zhang ZhengYou, and solve theinternal and external camera parameters, distortion parameters, rotation and translationvector. Then target image is preprocessed, Hough circle detection, corner detection,contour detection and Hough-chain code image pattern recognition algorithms. At the sametime, put forward to the arithmetic of center of multi-objective algorithm, combined cornerfeature the contour recognition algorithm, multi-target sub-block processing algorithm andHough-the chain code recognition algorithm. Aimed at the tracking dynamic workpiece onthe crawler, firstly, this method used Kalman filter to predict the position of subsequentmoving target, velocity and matching range, then used MeanShift algorithm based on HSVcolor space fusion to search in small range and match target, and finally the target positionwhich was achieved by the MeanShift algorithm was taken as the next frame’s Kalman filter input parameter to make the subsequent state get predictive capability, and executediteratively until finding target.The results of simulation show that the recognition algorithm can effectively solve thegeometric rules workpiece sorting problem, accurately calculate the center of theworkpiece, perimeter, area and edge parameters, in order to achieve accurate sortingagainst the static workpiece. For dynamic workpiece, tracking algorithm can effectivelysolve the tracking and positioning of the workpiece.